How rojld/agagi works
AGAGI is an empirical agent-and-world simulation research platform: researchers run evolutionary experiments in demanding simulated environments, assess whether cognition improves outcomes, and preserve findings as decision records and run data. Its dashboard exposes runs, comparisons, live activity, provenance, and the research record. The graph combines documented research capabilities with relationships visible in sampled source; several dashboard and backend subsystems are represented from the tree and README, so their internal wiring is not inferred.
Main entry points: biosphere.py Biosphere simulation · main.py FastAPI application · routes Product API routes · data_service.py Experiment data service
Components 22
Research workflow
- Research gates
- Experiment preflight
tools/experiment_preflight.py - Experiment runner
main_biosphere.py
Agents and worlds
- Mamba agent
src/agents/mamba_agent.py - Stoneage world
src/worlds/world_1_stoneage.py - Environment systems
src/environments - Biosphere simulation
src/environments/biosphere.py
Evolution and records
- Evolution engine
src/seed_ai/evolution.py - Genome mutation
src/seed_ai/mutation.py - Agent persistence
src/seed_ai/persistence.py - Experiment logging
src/graph_rag/async_logger.py - Experiment tracking
src/graph_rag/experiment_tracker.py - Decision record graph
src/graph_rag/database.py - Experiment visualization
src/visualization.py
API and services
- FastAPI application
backend/app/main.py - Product API routes
backend/app/routes - Experiment data service
backend/app/services/data_service.py - Run services
backend/app/services/runs_service.py
Research dashboard
- Research dashboard
frontend/src/App.tsx - Research views
frontend/src/components - Dashboard API client
frontend/src/api/client.ts
Other components
- Researcher
Connections 22
- Researcher → Experiment preflight: checks design
README.md - Researcher → Experiment runner: starts experiments
- Research gates → Experiment preflight: sets criteria
README.md - Experiment runner → Biosphere simulation: runs simulation
- Biosphere simulation → Genome mutation: mutates genomes
src/environments/biosphere.py - Biosphere simulation → Agent persistence: persists agents
src/environments/biosphere.py - Evolution engine → Genome mutation: applies mutations
src/seed_ai/evolution.py - Stoneage world → Environment systems: uses systems
src/worlds/world_1_stoneage.py - Stoneage world → Mamba agent: evaluates agents
src/worlds/world_1_stoneage.py - Biosphere simulation → Mamba agent: simulates agents
src/environments/biosphere.py - Experiment logging → Experiment tracking: logs experiments
src/graph_rag/async_logger.py - Researcher → Research dashboard: inspects findings
README.md - Research dashboard → Research views: renders views
frontend/src/App.tsx - Research views → Dashboard API client: requests data
frontend/src/components/ForageFunnelView.tsx - Dashboard API client → FastAPI application: calls API
- FastAPI application → Product API routes: mounts routers
backend/app/main.py - FastAPI application → Experiment data service: configures service
backend/app/main.py - Product API routes → Experiment data service: reads experiment data
backend/app/routes/academy.py - Product API routes → Run services: serves run operations
- Experiment data service → Experiment logging: reads experiment records
backend/app/services/data_service.py - Decision record graph → Research gates: stores decisions
README.md - Researcher → Decision record graph: curates conclusions
README.md